Jerame Huber. at the. May A u th o r... I... Department of Aeronautics and Astronautics May 16, 2003

Transcription

1 Noise Propagation Model for the Design of Weather Specific Noise Abatement Procedures by Jerame Huber Submitted to the Department of Aeronautics and Astronautics in partial fulfillment of the requirements for the degree of Master of Science in Aeronautics and Astronautics at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY May Massachusetts Institute of Technology All rights reserved. A u th o r... I Department of Aeronautics and Astronautics May 16, 2003 Certified by John-Paul B. Clarke Associate Professor of Aeronautics and Astronautics Thesis Supervisor I Accepted by Edward M. Greitzer H.N. Slater Professor of Aeronautics and Astronautics Chair, Committee on Graduate Students MASSACHUSETTS INSTITUTE SEP AERO LIBRARIES

2

3 Noise Propagation Model for the Design of Weather Specific Noise Abatement Procedures by J6r6me Huber Submitted to the Department of Aeronautics and Astronautics on May 16, 2003, in partial fulfillment of the requirements for the degree of Master of Science in Aeronautics and Astronautics Abstract This thesis presents an aircraft noise prediction simulation that incorporates actual weather in flight dynamics and noise propagation. The rapid prototyping simulation environment NOIse SIMulator (NOISIM) includes a sound propagation model based on a ray tracing algorithm that incorporates atmospheric and ground effects. The simulator uses standard weather profiles, terminal aircraft radar data and flight simulator data as input. NOISIM allows users to explore a wide array of flight procedures and weather conditions to determine the flight procedure that minimizes the noise impact in communities around airports. Two main applications of this tool are presented in this thesis: the design of a weather-specific noise abatement procedure and a statistical study of the effect of weather on average noise contours. The first case study explores the magnitude of the weather effects on the noise impact of a Boeing 767 in communities near Boston Logan Airport during takeoff. It also illustrates how the noise impact can be significantly reduced by changing the departure procedure to capitalize on changes in the weather. The second application is a statistical assessment of the impact of meteorology on annual average contours at major US airports. In this case we test the common assumption used in airport studies that weather effects on noise levels should average out over a year. Thesis Supervisor: John-Paul B. Clarke Title: Associate Professor of Aeronautics and Astronautics 3

4 4

5 Acknowledgments First I would like to thank my thesis supervisor Prof. John-Paul Clarke. My advisor greatly supported me in achieving my goals throughtout these three years. J-P, thank you for your knowledgeable advice regarding research and academic issues, as well as summer employment. With much room left for creativity, I learnt a lot while working with you! I appreciate all the great help that I got from our collaborators Steve Maloney and Rich DeLaura at MIT Lincoln Laboratory. Steve, Rich, your contribution was invaluable and very effective. It was great working with you. I address many thanks to my supervisor Kenneth Plotkin at Wyle Laboratories. Ken, thanks for keeping me inspired through a nice summer in DC. It is inspiring and encouraging to know you and the folks at the great lab of Wyle acoustics. Thanks you all the people in the International Center for Air Transportation (ICAT), especially in the room 220 for welcoming the French man. I leave my senior desk position in the lab with many great souvenirs. Manuel 'Chico', thanks for boosting the efficiency in the lab during a great summer Thanks to Alf Kohler for flying on the simulator. Salutations to my Canadian comrade Jonathan, Bruno, Yasmine and Diana for providing advice and keeping low noise levels of chatter. My special thanks go to Lilla, my dear friend, k6sz6n6m sz6pen, merci pour 6tre toujours la pour moi. Mad props to the kids in the World Student Mellen house. Berengere, Leslie, Marta, Claudio, Myriam, Alberto, and all, y'all provided precious encouragements through true friendship. Finally, I would like to express my appreciation for the constant encouragement that came from my parents from the other side of the Altantic ocean. P'pa et M'an, merci pour tout! 5

6 6

7 Contents 1 Introduction 1.1 Integrated Noise Model M ethodology Noise Metrics 2.1 Introduction to Sound The A-weighting Filter Effect of Duration Indices of Total Noise Exposure: 2.5 Noise Contours Day-Night Average Level NOISIM 3.1 Input Options Aircraft Performance Module Noise Source Module Noise Propagation Module Ray Acoustics Multiple Bounces Amplitude Variation Along Rays Atmospheric Absorption Excess Ground Attenuation Diffraction Into the Shadow Zone Summary of Propagation Effects

8 3.5 Output Options Noise M etrics Noise Contours Population Impacted by Noise Validation of NOISIM Effect of Weather on Noise Impact Weather Effect on Noise Propagation Through a Parametric Study Case Study: Actual Flights at Logan Airport Under Different Weather 48 5 Weather-Specific Noise Abatement Procedure 55 6 Impact of Meteorology on Average Contours Noise Propagation Algorithm Preparation of Meteorological Data Airport M odeling Noise Contours Contours with Idealized Atmospheres Hourly and Monthly Contours in Real Weather Contours Averaged Over One Year Conclusions Sum m ary Implications for Air Traffic Control and Future Research A Derivation of the Ray-Tracing Equations 71 A.1 The General Ray-Tracing Equations A.2 Propagation in Stratified Atmosphere B Derivation of the Blokhintzev Invariant for Wave Amplitude 77 C Derivation of the Equations for Atmospheric Absorption 81 8

9 D Derivation of the Equations for Excess Ground Attenuation D.1 Theory of Source-Receiver Geometry D.2 The Ground Absorption Model D.3 Turbulence E Derivation of the Diffraction Theory F Properties of the Atmosphere F.1 Tem perature F.2 Relative Hum idity F.3 W ind F.4 International Standard Atmosphere G Impact of Meteorology on Average Contours

10 10

11 List of Figures 2-1 Relative response of the A-weighting filter, from Ref. [19] Noise footprint in db SEL generated by NOISIM - Departure of a B767 from Logan 4R on May 25th, Structure of the Noise Simulator NOISIM Typical turbomachinery sound pressure level polar directivity patterns, from R ef. [12] Boeing 767 flying through microphone array at the Wallops Flight Test Facility Directivity for Boeing at 100ft and 2000Hz. The noise intensity is plotted versus azimuth angle. Each series has a constant polar angle Effects in the noise propagation Sketch of ray tube, from Ref. [17] Source-receiver geometry Curved ray path above plane boundary, from Ref. [2] Map of Population Density in Greater Boston Area - USCB data for the year dba Contour Areas for a Static Jet Engine Source at 1000m Above G round Simulated trajectory for existing departure procedure - runway 4R, Boston Logan airport Areas of Contours Generated for Departure Procedure Under Different Temperature and Humidity Conditions

12 4-4 Areas of Contours Generated for Departure Procedure Under Increasing Cross Wind Magnitude and Shear All SMA Departure Tracks Flown During the Month of May 2002 out of Runway 4R at Boston Logan Intl' Airport Weather Profile for Winter Case Weather Profile for Summer Case SEL noise contours for Winter Case Generated by NOISIM - Simulated Standard Noise Abatement Procedure at Boston Logan Runway 4R SEL noise contours for Summer Case generated by NOISIM - Simulated Standard Noise Abatement Procedure at Boston Logan Runway 4R Simulated trajectory for existing departure procedure - runway 4R, Boston Logan airport SEL noise contours for existing departure procedure Simulated trajectory for weather-specific departure procedure - runway 4R, Boston Logan airport SEL noise contours for weather-specific departure procedure Contours at 65dB for hourly operations sample from January 96 at ORD (a): Uniform Atmosphere (b): Standard Atmosphere Contours at 65dB Leq averaged over the month of January 96 at ORD airport Standard deviation between hourly contours over January 96 at ORD airp ort Annual Lda contours at (a): CPR (b): RNO Annual averages contours areas for 7 US airports Concept of an advisory system for ATC to achieve weather-specific noise abatement procedures A-1 Concept of a ray path B-1 Sketch of ray tube, from Ref. [17]

13 C-1 Attenuation coefficient a as a function of frequency for a temperature T = 204C and relative humidity as parameter, from Ref. [13] D-1 Source-receiver geometry E-1 Curved ray path above plane boundary, from Ref. [2] F-i Refraction of sound waves by wind gradients following a power-law profile, from Ref. [20] F-2 Exponent for power-law wind profile, as function of surface roughness 98 G-i Annual Ldn contours at (a): CPR (b): IAH G-2 Annual Lda contours at (a): LAS (b): LGA G-3 Annual Lda contours at (a): MSY (b): ORD G-4 Annual Ldn contours at RNO

14 14

15 List of Tables 3.1 Measured and Predicted Noise at Stations Near Logan Airport Deviations Between Predictions and Measurements at Noise Stations Near Logan Airport Population impacted by single B767 departure from Runway 4R in winter and summer Population impacted by noise (in 10 dba bins) for existing and weatherspecific departures from runway 4R

16 16

17 Chapter 1 Introduction Airports are simultaneously facing increased demand for their services as a result of economic growth and increased community opposition to expansion because of environmental concerns. Chief among these environmental concerns is the impact of aircraft noise on local communities. In response to community concerns about noise, many airport authorities, in collaboration with the Federal Aviation Administration (FAA), either soundproof homes or implement noise abatement procedures based on the output of noise models such as the Integrated Noise Model (INM). While it is well known that weather affects both the performance of aircraft and the propagation of noise [22], many noise models do not incorporate the effects of weather on either aircraft performance or noise propagation. Those models that do, only do so in an approximate manner because of computational limitations. As a result, noise procedures are often developed for a single weather scenario that is thought to be representative of average annual conditions. In reality, daily variations in weather can cause significant changes in the noise impact and thus the 'best' noise abatement procedure to use on any given day can vary significantly. 1.1 Integrated Noise Model The existing standard for determining the impact of aircraft noise at and around airport is the Integrated Noise Model (INM). This model, developed by the FAA, 17

18 estimates the average noise exposure at airports over extended periods of time such as a day or a year. INM is designed as a planning tool to help airport operators decide on the appropriate runway configurations to use given a specific mix of aircraft types operating from that airport. In order to predict the average impact, INM uses typical approach and departure profiles, empirical in-flight noise data, a user specified mix of aircraft types, and a user specified number of operations. Although the fidelity of INM is sufficient for planning purposes, it has a number of limitations. Among these limitations is the fact that INM does not account for real weather conditions in the modeling of aircraft performance and noise propagation. INM also features empirical in-flight noise data, so the noise impact of trajectories that are outside of the range of trajectories stored in the INM data base must be extrapolated. The trajectory of an aircraft can only be described by a limited number of straight line segments [6]. The need therefore exist for a methodology to evaluate the impact of aircraft noise with maximum accuracy while accounting for actual weather conditions. To accomplish this goal, the design methodology must include aircraft and noise modeling capabilities that can be used in combination to evaluate the noise impact of single approaches and departures. 1.2 Methodology In this thesis, we explore the impact of weather on the noise exposure around airports, and illustrate how knowledge of this noise impact might be used in air traffic control to select departure and arrival trajectories that have the lowest noise impact for given weather conditions. While this represents a radical change in the way noise is considered in the daily application of noise abatement procedures, this approach provides a novel way to tailor flight operations at airports to the needs of the community, in that it allows the noise impact to be consistent irrespective of weather. The noise impact is evaluated using the latest version of the Massachusetts Institute of Technology (MIT) NOIse SIMulator (NOISIM), a rapid prototyping simula- 18

19 tion environment that allows users to explore a wide array of flight procedures and weather conditions to determine the flight procedure that minimizes the noise impact in communities around airports. NOISIM incorporates atmospheric and ground effects, using standard weather profiles, terminal aircraft radar data and flight simulator data as input. The structure of the thesis is as follows. Background on the noise metrics that are used in the thesis, and the rationale for selecting these metrics, is provided in Chapter 2. The details of NOISIM are presented in Chapter 3. Input options, aircraft performance and noise source are described first, followed by an overview of the noise propagation algorithm based on a ray-tracing technique, and featuring amplitude variation, atmospheric absorption, excess ground attenuation and diffraction in the shadow zone. The effects of weather on surface noise are presented in Chapter 4. Parametric analyses and case studies illustrate changes that can be expected in the noise impact for different weather conditions. A study case at Boston Logan is presented in Chapter 5. This analysis illustrates how the methodology can be used to evaluate, compare noise abatement options and achieve benefits with a noise abatement departure procedure that is specifically tailored to the weather conditions. A statistical study at seven major airports in the United States is presented in Chapter 6. It shows the capability of the model to survey the impact of meteorology on annual average contours at various locations. A summary of the results of the work, conclusions based on these results, suggestions for future work and implications for air traffic control are presented in Chapter 7. 19

20 20

21 Chapter 2 Noise Metrics A description of the measures and metrics used in this thesis is presented below. 2.1 Introduction to Sound The sound source sets the nearest particle of air into vibration through which acoustic energy is transmitted to the surrounding air. The motion of air particles about their equilibrium position produces a local compression followed by a local rarefaction and so on. The instantaneous value of the fluctuating pressure disturbance on the ambient pressure is called the sound pressure and is given the symbol p(r, t) for a spherical wave. The human ear responds to sound intensity, which is related to sound pressure in free field by 2 I =- (2.1) pooc There are a number of metrics that are used to measure the impact of noise at a particular location. Some metrics measure the peak instantaneous noise level while others have some correction for the effects of duration. The most basic metric is the sound pressure level (SPL) defined by 21

22 SPL = 10 log 10 O (2.2) Pe0 2.2 The A-weighting Filter In addition to linear response which allows the measurement of sound levels, frequency weighting filters are introduced to reflect the loudness caused by a complex sound. The human ear reacts differently to sounds of distinct frequencies. The filters make a correction for each sound pressure level according to frequency. Four weighting filters have been standardized, designated A, B, C and D. The A-weighted sound level is the most widely employed for assessing sounds of all levels. It was developed to account for the response of the human ear to noise at different frequencies. Through this filter the contributions from the low order harmonics are considerably reduced as presented in Figure Relative response db Frequency, 1i Figure 2-1: Relative response of the A-weighting filter, from Ref. [19] The overall A-weighted sound level (dba) is obtained for a given tertsband sound pressure level spectrum by: L4 = 10 logio E 10 1 dba, (2.3) where LA(i) = SPL(i) + AL, 1 (i) is the corrected band level. The peak dba at the receiver is the instantaneous maximum LA observed during the event. 22

23 2.3 Effect of Duration The equivalent A-weighted sound level is an important descriptor for the subjective loudness of an airplane flyover noise. It is obtained by continuous integration for a specified time period T. LAeq,T = 10 log 1 0 j 10f dt] dba. (2.4) To include the effect of duration and at the same time eliminate the influence of the measurement duration T, the sound exposure level (SEL) is employed. LAE = 10log To dt dba, (2.5) where LAE is the symbol for sound exposure level and Ti is the reference time of one second. 2.4 Indices of Total Noise Exposure: Day-Night Average Level The day-night average level Ldn or DNL is the equivalent A-weighted sound level integrated on the basis of the squared pressures over a 24-hour period. The noise levels occurring at nighttime are increased by 10 db in order to account for the annoyance caused by noise heard during the nighttime hours from 22:00 to 07:00. This metric was developed in the United States as a measure for environmental noise. DNL can be determined by continuous integration of A-weighted sound level for a second (24 hour) period. - 1 f Ldn = 10 log f8600 I w(t)10 10V dt], (2.6) where w is the weighting factor for the time of the day. From 07:00 to 22:00, w = 1; from 22:00 to 07:00, w = 10. The DNL metric combines four major factors in noise annoyance into a single 23

24 index. Three of these factors, loudness, duration and number, are combined in the Leq. The fourth, time of day, is incorporated through the nighttime penalty in DNL. 2.5 Noise Contours The noise heard on the ground near an airport can be presented by plotting lines of constant noise levels. A single-event noise contour or footprint gives the noise level at any particular point on the ground expressed in one of the single number noise descriptors. In the joint Wyle-MIT study of annual averages in Chapter 6, descriptors of total exposure DNL were plotted as noise footprints. In the case study for departure from Boston Logan 4R presented in Chapter 5, the noise impact was plotted in peak A-weighted sound pressure levels (peak dba), and in SEL to show effects of duration. Figure 2-2 shows a noise footprint predicted by the model at MIT, for a Boeing 767 taking off from Boston Logan runway 4R. Contours are identified with colors. Numbers on contour lines indicate the minimum value of noise levels in db SEL inside the closed line. 24

25 Figure 2-2: Noise footprint in db SEL generated by NOISIM - Departure of a B767 from Logan 4R on May 25th,

26 26

27 Chapter 3 NOISIM The structure of NOISIM is shown in Figure 3-1. As the figure shows, the user inputs the weather conditions and either the procedure that was used by the aircraft or the trajectory that was flown by the aircraft. If the procedure in input, the procedure and the weather are used in the aircraft performance module to determine the aircraft trajectory. The trajectory is then used in the noise source module to determine the noise source characteristics, and then subsequently used in the noise propagation module, with the noise source characteristics and weather, to determine the noise impact on the ground. 3.1 Input Options Weather is specified as a function of altitude, and is typically derived from balloon soundings. Balloon soundings provide the pressure and derived altitude, temperature, the dew point temperature, and wind field, in magnitude and direction for a vertical profile of the atmosphere. Combined with the temperature, the dew point gives the humidity in the air. The data from balloon soundings is interpolated to layers of equal thickness and a horizontally homogenous atmosphere is assumed. The weather profile is eventually entered into the model as a table of altitude (meter), pressure (kpa), temperature (degree Kelvin), absolute humidity (%), components of wind in the east and north directions (m.s 1 ), and air density (kg.m 3 ). 27

28 NOISIM II Figure 3-1: Structure of the Noise Simulator NOISIM Aircraft procedures are specified as they would be to a pilot. That is, a procedure is specified in terms of throttle settings/targets as a function of altitude and speed, and speed settings/targets as a function of altitude. If the procedure is input, the trajectory is determined using the aircraft performance module described below. The aircraft trajectory may also be input directly. The aircraft trajectory is typically derived from radar data. In prototype implementation, radar data was obtained from the SMA (Surface Movement Advisor) program. This data included the position and velocity for each aircraft in time increments of 15 seconds. 28

29 3.2 Aircraft Performance Module The aircraft performance module in NOISIM is a point-mass numerical simulation model based on the drag polar-the curve that describe the drag coefficient as a function of lift coefficient. The module thus determines the position, speed, attitude and thrust as a function of time by integrating the forces and thus acceleration on a point-mass that is equivalent in weight to the weight of the aircraft. This information is used with a look-up table to determine the operating condition of the engine as a function of time as represented by throttle setting parameters such as the variable N1, a relative measure of the fan rotation speed. 3.3 Noise Source Module An acoustical source is defined by its noise spectrum and directivity. That is, the distribution of intensity in terms of both frequency and direction, where the direction is defined by two angles relative to the body axis of the aircraft: the polar angle and the azimuth. As shown in Figure 3-2, the noise directivity of a turbofan engine is defined by the polar angle. This is understandable given the fact that the engine is symmetric around its axis of rotation. The aircraft source noise directivity, however, represents the pressure disturbance created by the entire aircraft. In this case, the source is not axi-symmetric, thus the noise directivity must also be described in terms of the azimuth. While it is possible to generate noise spectrums for commercial airliners such as the Boeing 767 with the appropriate engine specifications and a noise prediction program such as ANOPP [NASA], a more accurate description of the aircraft source noise spectrum and directivity may be derived from flight test data. To that end, noise data from a recent flight test of a Boeing 767 at the Wallops Flight Test Facility in Virginia were used to derive a noise spectrum and directivity. At the tests, the Boeing 767 flew through a microphone array with microphones mounted on cranes as depicted in Figure 3-3. Thus, it was possible to create a complete hemisphere that describes the 29

30 Inlet _ Flow Flow E Rad. rtone SPL,,,,. - Broadband Polar angle, 8, deg Figure 3-2: Typical turbomachinery sound pressure level polar directivity patterns, from Ref. [12] noise below a Boeing 767 aircraft as a function of throttle setting and speed. Figure 3-4 shows the directivity for the Boeing at a distance of 100ft and a frequency of 2000Hz. Directivity patterns are defined for the center frequency of each third-octave band, and are corrected for the effects of interference. 3.4 Noise Propagation Module The core of the noise propagation module is the ray-tracing algorithm that is described below. The propagation of sound is modeled using far-field acoustics, assuming that the sound wave is spreading spherically. The ground is assumed to be flat and described by its specific flow resistance. The receiver height is assumed to be four feet above the ground. The sound propagation effects are shown in Figure 3-5. In the first step, the ray paths from the source to the ground are determined by numerically integrating of the ray-tracing equations. The ray tracing algorithm computes sound paths by integrating the effects of the local wind and temperature gradients. The 30

31 Figure 3-3: Boeing 767 flying through microphone array at the Wallops Flight Test Facility algorithm accounts for multiple bouncing. That is it follows rays that may bounce several times at the ground. Once all the ray paths have been determined, the propagations effects are computed. The wave amplitude, which is governed by spreading and focusing effects, is determined by measuring the intensity in ray tubes defined by three adjacent rays. Atmosphere absorption of sound is computed with empirical formulas that describe the absorption as a function of frequency, temperature and humidity. Excess ground attenuation is computed at each point where direct rays and reflected rays interfere with each other at the height of the receiver. Finally the model provides a realistic transition between illuminated and shadow zones, the zones in which no ray can penetrate. As opposed to illuminated zones where the geometrical approach is valid and accurate, shadow zones must be modeled using diffraction theory Ray Acoustics The atmosphere is modeled as a horizontally stratified medium. Rays are propagated using the equations developed by NASA [23] to calculate ray paths in a stratified atmosphere. The derivation of the ray-tracing equations can be found in Appendix A in this thesis. These equations are equivalent to the equations developed by Pierce 31

32 B Hemisphere Data - Pass Series Il "20"= ANSIIlSO Band Hz; Individual Series Represent Unique Polar Angles Symmetric Spectrum s Azimuth Angle (degrees) Figure 3-4: Directivity for Boeing at 100ft and 2000Hz. The noise intensity is plotted versus azimuth angle. Each series has a constant polar angle. source Ray tracing Tube area Atmnosphere absorption Excess Ground D~gkac n nattenuation Mulple Figure 3-5: Effects in the noise propagation [17] and were initially developed to extrapolate sonic boom signatures. The ray path vector X(i) and wavefront unit normal vector N(i) are determined from the equations: X(i-+1) = X(i)+AX(i) N(i + 1) = N(i) + AN(i) AX(i) ANi(i) = (ao(i)n(i) + Vo(i))At Ni1(i) N (i) AN 2 (i) =F(i) N 2 (i)n 3 (i) x At A N 3 (i)) -N (i - N21 32

33 a VO 02o + ao F(i) = N1(i) 1 (i) + N 2 (i) (i) + (i), (3.1) Oz Oz az where At is the time increment between points. The typical time increment is 0.05 second. In the vicinity of the receiver and ground, the time increment is reduced so that each iteration traces a path of length half the distance between the receiver and the ground. This choice improves the precision of the location of ray impact at receiver and ground. ray: Ray-specific arrays are used in the algorthm to store the following data for each " the impact location (i.e. where the ray crosses the receiver or the ground). * the time and the length of ray path at impact. " the angle of reflection at ground. Ray tracing is stopped when one of the following conditions are met: * The ray has reached the ground twice. This means it has reached the receiver, reflected on the ground and did a full bounce. " The maximum number of time steps (9999) is exceeded. This limitation of the number of time increments cuts the computation of upward propagating rays that do not hit the ground. " The ray propagates outside the radius of tracing. The values of time, path length and angle are projected on a global grid that represents the portion of land for which noise is predicted. Propagation effect modules such as atmospheric absorption, excess ground attenuation and diffraction process grids instead of individual rays. Grids are an alternative to lengthy ray data tables. Processing rays would imply a classical shooting method that searches pairs of direct and reflected rays that both reach the same observer. The benefits of the gridding method are a much shorter calculation time and a simpler algorithm. The fortran routine Zgrid is used in the model to grid values. 33

34 3.4.2 Multiple Bounces In the illuminated zone, a ray will typically first reach the receiver along a path directly between the source and the receiver. When the receiver is above ground, other rays that are reflected at the ground may propagate upwards. If conditions are such that the path is convex, the reflected ray will bend back toward and hit the ground again, as depicted on the right-hand side in Figure 3-5. If the ray hits the ground inside the grid, the ray is reflected at the ground and propagated as before. Since the atmosphere is horizontally stratified, the paths of the rays between bounces will be the same. Thus, the first bounce is traced and all subsequent bounces are then extrapolated until the reflection at the ground is off the defined map. Effects such as atmospheric absorption and excess ground attenuation are included at each bounce. Pressure is gridded according to the number of bounces. Embleton [11] estimates that the maximum contribution of multiple bounces is 2.2 db Amplitude Variation Along Rays Each point on the wavefront has an amplitude and an associated phase. When the rays propagate in slowly varying medium, the phase is directly related to the time it takes for the wave to travel from the source to the receiver. The amplitude of the sound is computed using ray tubes. A ray tube consists of all rays passing through a small area A(xo) at coordinate xo. When the ray tube reaches the position x, its cross-sectional area is A(x). Figure 3-6 illustrates this concept. The intensity along ray tubes is computed using the Blokhintzev approach. The Blokhintzev invariant given in Equation 3.2 is constant along any given infinitesimal ray tube of variable cross-section area A: (1 - VTp 2 = constant. (3.2) (1 - V -Vr)pc2 For the simple case of rectilinear propagation in uniform still atmosphere, the invariant gives the same results as the geometrical spreading of the intensity, I oc 1/r2, r being the distance from the source. The area of cross-sections A indeed behaves as 34

35 A(xo) Figure 3-6: Sketch of ray tube, from Ref. [17] r2 in the rectilinear propagation case. This case of uniform atmosphere was used as benchmark to test numerical precision. The details of the derivation to obtain the Blokhintzev invariant is presented in Appendix B. The elementary tube taken in the calculation consists of three rays. For each ray traced by the regular ray tracing routine, the amplitude routine traces two other rays and establishes the ratio of areas. The cross-section of a three-ray tube is a triangle. One ray impacts first on the ground and the impact is chosen as the first of the three vertices. The other two vertices of the cross-section are calculated by interpolation so as the three vertices are on the same wavefront, with equal phase. The square pressures are established at both ends of the ray tube. The transmission loss is the intensity difference in db between the source and receiver along a given ray due to the change of area of the ray tube. The transmission loss is given by T Lss 10log 10 ( receiver TLoss = 10 lg P 2 source [(1 - V - VT) PC 2 ]receiver [IVray A]ource [(1 - V. VT)PC 2 ],source [Vray Arecever J The transmission loss is calculated for each ray. Given the impact point (x, y) at the receiver, the transmission loss is then mapped on the global grid. Points outside the illuminated zone are considered in the shadow zone and are treated by 35

36 the diffraction routine presented below Atmospheric Absorption Atmospheric absorption is the process by which sound energy is dissipated as a sound wave travels through the atmosphere. This form of attenuation is due primarily to the fact that air exhibits some degree of internal friction. The vibrational relaxation of oxygen and nitrogen molecules is the major phenomenon responsible for this energy dissipation. Molecular absorption converts a small fraction of the energy of the sound wave into internal modes of vibration of the oxygen and nitrogen molecules. The total absorption is determined using the relationships developed by Bass, Sutherland and Zuckerwar [13]. Detailed equations can be found in Appendix C. The attenuation coefficient a 3 in db/100m for each layer j is given by T 1/2 a = p, - F x 10" (- ) PSO TO + T -5/2 e / e-3352/ ) + - ~ ~ xf.0251f ~ F~.16 TO F, FFr'o Fr,N + F2 r,n where po is the reference value of atmospheric pressure. To = K is the reference atmospheric temperature, T is the atmospheric temperature in K. Experimental measurements give estimates for the relaxation frequencies of molecular oxygen Fr,o and nitrogen Fr,N scaled by atmospheric pressure. These relaxation frequencies are functions of temperature and humidity [13]. atmospheric pressure. F is the frequency in Hz scaled by To avoid lengthy calculations, the atmospheric absorption is computed for the straight path between source and receiver, rather than for the slightly longer curved path. The approximation of straight path for absorption introduces an error less than one percent of the total attenuation. As a result, the elementary path lengths sj are identical in all layers j, and the formula for the total atmospheric attenuation is given 36

37 by ATT = Slay. (3.3) layers Excess Ground Attenuation An observer near the ground can receive sound directly from the source or from reflections at the ground. When sound is reflected at the ground, the amplitude of the wave changes. The phase of the reflected ray differs from the direct ray because the path of propagation is different, as illustrated in Figure 3-7. The ground attenuation effect is computed at the height hr of the receiver by using the properties of the locally reacting ground, the path lengths and phase differences of sound rays. The interference patterns between direct and reflected rays is determined using the methodology described by Chessell [4]. In this methodology, the ground is characterized by a single parameter, the specific flow resistance per unit thickness o-. The terrain geometry is considered flat. Figure 3-7 illustrates the source-receiver geometry used in the development of the theory. The use of such model was originally proposed by Delany and Bazley [9]. The two scientists considered plane wave reflections. Chessell extended their work with the single flow resistance parameter but to study the reflection of a spherical wave at the boundary. His method gives results that are in good agreement with experimental data, showing that real soil surfaces can be represented by the simpler local reaction boundary condition assumption. The time-averaged excess attenuation over third-octave bands is given by Chessell [4]: <Ae > = 101log[1+ rqil2 2 p1q cos(r;ar/a-+ 06) -02A~ Jx + Qj Isin e 2 37

38 S hrr r R k * 2h,. Z, k, Z 2, k 2 Figure 3-7: Source-receiver geometry y = 2irAf /2fi r7 = 27r[1 + (Af/2fi) 2 ] 1 / 2. (3.4) Q is called the image source strength, it reflects the ratio of magnitudes of reflected and direct waves and phase difference. Ar = r2- r 1 is the path length difference between the direct and reflected ray. The standard deviation of the path difference fluctuations ao, is introduced to correct the attenuation for turbulence in the atmosphere. The effect of turbulence is to create incoherence in the interference pattern, and thus reduce the maximum excess attenuation and cause a shift in the peak excess attenuation to lower frequencies. The effect is smaller for short ranges and low frequencies. The detailed derivation of the excess ground attenuation is presented in Appendix D. The routine called ega for excess ground attenuation works on grids. That is, instead of calculating the interference for each direct and reflected rays at the same receiver. The routine ega computes the interference between the 'direct-ray grid' and the 'reflected-ray grid'. Corrections for phase differences are performed at the third-octave band center frequencies. This results in reduced computation time and algorithm simplicity Diffraction Into the Shadow Zone In geometrical acoustics, shadow zones are areas where no ray can penetrate. The upward bending of rays can create shadow zones as illustrated in Figure 3-8. The re- 38

39 gion is determined by the limiting ray which just strikes the ground surface. However an observer would still perceive low levels of noise in the shadow zone. Where the geometrical solution described in Section appears to be invalid in the shadow zone, we compute a solution solving the problem using diffraction theory. The development of the diffraction module is based on the work of A. Berry and G.A. Daigle [2] and can be found in more details in Appendix E. Z R lirnitng ray (2hsa sr Figure 3-8: Curved ray path above plane boundary, from Ref. [2] A significant amount of sound energy penetrates into the shadow region via a creeping wave. Pierce [17] showed that the creeping wave travels near the ground inside the shadow region and sheds energy upward during propagation. The creeping wave is strongest as it enters the shadow region and becomes weaker with distance. Pierce's work on soft and hard boundaries was extended to the general case of surfaces having finite impedance by Berry and Daigle [2]. The sound field is described by the pressure p(r, z) as function of range and altitude. The creeping wave solution is expressed in terms of residue series that give the pressure as,reir/6 (1)kAi[bn - (hs/l)e 2 4x/ 3 ]Ai[bn - (z/)e 2 i/ 3 p(r, z) = 1: He knr [Ai'(b )] 2 - bn[ai(ba)] 2, (3.5) where b- = re2ix/3 = (k2 - k )12,2ix/3 (3.6) 39

40 are the zeros of the expression Ai'(bn) + q[eix/ 3 ]Ai(bn) = 0. (3.7) T = The abbreviations in Equations 3.5, 3.6 and 3.7 are q = [ikoplc]/z, 1 = (R/2k2) (k 2 - k2)1 2, and ko = w/c(o). R is the radius of curvature of the limiting ray as shown in Figure 3-8. The limiting ray is tangent to the ground, thus values for wind and temperature gradients used to determine R are picked in the lower atmospheric layer Om to 10m above the ground. The residue series solution is valid near the boundary and deep in the shadow zone. When the calculation is close to the edge of the shadow zone, more terms are computed to make the series converge. Generally a couple of terms are sufficient to obtain a good estimation of the series in the shadow zone. Finding the roots b" of Equation 3.7 is particularly difficult and time-consuming. Preference is therefore given in the algorithm to approximations of pressure-release surface (Z. -+ 0) and rigid surface (Zg -> oo) Summary of Propagation Effects The noise level at a given location is computed using the relationships described below depending on whether that location is in an illuminated or a shadow region. SPLiulum = SPLSourceNoise + SPLAtmosphericAbsorption + SPLTransmissionLoss + SPLGroundAttenuation SPLshadaow = SPLSourceNoise + SPLAtmosphericAbsorption (3.8) 40

41 + SPLDif f ractionsolution (3.9) 3.5 Output Options The sound propagation module outputs the noise impact in the form of a grid that represents the ground. A typical grid with dimensions of 32 x 32km has 40,000 square cells that are 160 x 160m in size Noise Metrics Because the noise spectrum is defined in term of third-octave bands, the noise reaching each cell will be initially given in third-octave bands. However, the noise propagation module outputs noise in terms of the peak dba and SEL in each cell. Thus, for a given tertsband sound pressure level spectrum, the overall A-weighted sound level is determined using the expression: LA = 10 log E dba, (3.10) where LA(i) = SPL(i) + ALA(i) is the corrected band level. The sound exposure level (SEL) is determined using the expression: LAE = 10log[ 10+ 1dt dba, (3.11) where LAE is the symbol for sound exposure level and T is the reference time of one second Noise Contours Noise contours, a graphical representation of the noise impact on the ground, are produced using the MATLAB mapping program mmap. The contours are created by plotting the data contained in the grid that is output by the sound propagation 41

42 module on a map of the airport and its environs Population Impacted by Noise The population impacted by aircraft noise at specific levels is determined using United States Census Bureau (USCB) population density data for the year Figure 3-9 shows the population density in the greater Boston area near Logan airport according to census Population Density, Boston, MA Population per sq. mi y ~ ~ W U N S Source: US Census Logan Airport 0 4 SMiles Figure 3-9: Map of Population Density in Greater Boston Area - USCB data for the year 2002 Specifically, the impacted population is determined using the geographic information system (GIS) software program ArcView by ESRI. This is accomplished as follows. The data from two different data sets are plotted on the same Latitude/Longitude coordinate system. The first data set gives the population density based on the 2000 Census data at block group level. The second data set gives the noise in each of the 160 x 160m cells in the grid produced by the noise propagation module. Data from both attribute tables are then combined into a single table 42

43 using geo-processing. The population impacted by the noise is then calculated by multiplying the area being impacted and the population density. 3.6 Validation of NOISIM The aircraft performance module has been validated against both level-d simulator performance data and flight data recorder data. The noise source module was developed using flight test data so is therefore validated against measured data. Each routine in the noise propagation module is validated using the benchmark problems, data and results reported by the respective authors in the papers that describe the theories and their validation. The transmission loss is validated for the case of a uniform atmosphere as the loss in that case is solely due to spherical spreading. The accuracy of the predicted noise levels was validated with data from four noise monitoring stations around Boston Logan Airport. This validation was performed by predicting the noise at these four stations for all trajectories that were flown over a period of four days beginning on May 25th, 2002 and then comparing the predicted noise levels with the measured noise levels. Table 3.1 shows the mean measured and predicted noise levels for the departures, while Table 3.2 presents the error distribution for the same cases. As can be seen, the noise in terms of dba is slightly over-predicted, while the noise in terms of SEL is very close to the measurements levels. It should also be noted that the variation in noise level with weather is highly correlated. Table 3.1: Measured and Predicted Noise at Stations Near Logan Airport Noise Station Measured dba Predicted dba Measured SEL Predicted SEL

44 Noise Station Delta dba Delta SEL Table 3.2: Deviations Between Predictions and Measurements at Noise Stations Near Logan Airport 44

45 Chapter 4 Effect of Weather on Noise Impact This chapter aims at showing the effects of weather on surface noise. Parametric studies are run with idealized weather profiles. Real weather and radar flight data is used in a case study for actual departures from runway 4R at Boston Logan airport. 4.1 Weather Effect on Noise Propagation Through a Parametric Study The study of a single emission from a static point source representing the aircraft allows to draw some conclusions regarding the noise propagation. However, it does not reflect the effects of weather on the noise propagated during a whole departure or arrival flight. Conclusions for a static source are presented in Figure 4-1 where the areas of all contours are shown to increase when air humidity increases around standard conditions for temperature. Around standard relative humidity of 70%, the largest contours are obtained for temperatures near 100C and they shrink away from this average temperature. In reality, full aircraft footprints are studied as a succession of different source emissions in time, with weather conditions varying in altitude. In this section, we compare the areas of noise contours in populated zones, caused by a single departure 45

46 Figure 4-1: dba Contour Areas for a Static Jet Engine Source at 1000m Above Ground flight in various weather conditions. The effect of weather is studied by using idealized weather profiles. Artificial profiles are generated with desired values for temperature, humidity and wind magnitude and direction. Analyses isolate the effects of weather variables on noise propagation. The variables identified in the weather profiles are temperature, temperature gradient, humidity, wind magnitude shear and wind direction shear. A standard departure procedure out of Boston Logan 4R was simulated as shown in Figure 4-2. It was flown as an existing noise abatement procedure further described in Section 4.2. With this same flight, different weather profiles are input into NOISIM to assess the noise impact. The area of each noise contour is calculated and results are reported in charts. Contours were considered near populated areas since the flight was simulated until coastal lines were out of reach. This is the reason why contours of high intensity above 70 db SEL are over-represented in the charts in terms of absolute total area. The study reflects the effect of weather on noise contours in the vicinity of the airport. General conclusions can be drawn from the observation of both static source cases and parametric studies on the single-event departure case. Temperature and humidity values play a key role in atmospheric absorption. Fig- 46

47 Figure 4-2: Simulated trajectory for existing departure procedure - runway 4R, Boston Logan airport ure 4-3 shows contour areas for different values of temperature and relative humidity, plus one case of temperature inversion with an opposite standard temperature lapse, namely 6.50C/km. Significant deviations are observed between a cold dry day, and a warm humid day. A cold dry day implies less absorption during the propagation of sound than when the air is warm and humid. Figure 4-3 also illustrates that a temperature inversion affects noise contours areas, especially on contours with lower noise levels. The contours of highest noise levels do increase in size when temperature rises to 35 0 C, they have comparable sizes in the other weather cases. The contours of higher intensity are mostly located under the aircraft, and whereas temperature and wind gradients do not affect significantly these high noise levels, atmospheric absorption can alter them in a significant manner. Temperature gradients, wind magnitude and wind direction shear cause sound refraction in the atmosphere. The sound therefore propagates in different directions when one of these parameters varies. The high-intensity contours above 85 db SEL do not generally change as the sound propagates nearly perpendicular to the wind vector which changes in magnitude and direction. At the same time, lower intensity contours do vary both in size and location when such changes occur in the wind field 47

48 Contours Areas for Different Conditions of Temperature And Relative Humidity ' - *- T=15C, RH=70% -E-T=25C, RH=70% - T=35C, RH=70% -+-T=15C, RH=10% Temp Inversion T=15C RH=70% - std lapse db SEL Figure 4-3: Areas of Contours Generated for Departure Procedure Under Different Temperature and Humidity Conditions because the sound propagation at shallower angles and over greater distances is more sensitive to gradients. Figure 4-4 shows that cross wind alters contours below 75 db SEL. The values of the wind are set at the top of the wind boundary layer. The model used for the wind boundary layer is found in Appendix F. Changes in head wind have generally less of an effect in terms of contours areas and only full contours under 65 db SEL are modified. Wind direction shears affect contours significantly below 85 db SEL. Noise intensity is generally spread over greater surfaces when wind direction shear is present, and contours at 70 db and 75 db SEL increase in size. 4.2 Case Study: Actual Flights at Logan Airport Under Different Weather Jet aircraft departing from Runway 4R are required to perform a noise abatement procedure designed to reduce the noise impact in residential communities. In the existing procedure, the pilot maintains the runway heading of 040 until the aircraft is 4 nautical miles from the Distance Measurement Equipment (DME) beacon located at the airport. At this point the pilot changes the heading to 090 and flies toward the Atlantic Ocean. The net effect of the procedure is that aircraft execute a turn to 48

49 Contour Areas With Increasing Wind Magnitude and Shear Wind Value at Top of Boundary Layer Wind 0 m/s - U - Wind = 4 m/s = 8 mis s db SEL Figure 4-4: Areas of Contours Generated for Departure Procedure Under Increasing Cross Wind Magnitude and Shear the right after passing abeam of Nahant regardless of the weather conditions. Figure 4-5 shows the tracks of all departure procedures flown out of runway 4R during the month of May 2002, recorded by radar and provided by the SMA program. 30,r SMA Fqht Tratis 424N 24.00' / ( (7 Figure 4-5: All SMA Departure Tracks Flown During the Month of May 2002 out of Runway 4R at Boston Logan Intl' Airport To illustrate the effect of weather on the noise impact, the noise impact was calculated for actual departures during representative weather conditions in winter (winter case) and summer (summer case). In these two cases, illustrated in Figure 4-6 and 49

50 Figure 4-7, the winds are similar, but the temperature and humidity are significantly different. The meteorological data used in both cases was obtained from the National Weather Service Rawinsonde Observation (RAOB) from Chatham MA (CHH). The position and velocity of each aircraft in 15 seconds increments was obtained from radar data provided by the SMA program. Figure 4-6: Weather Profile for Winter Case Figure 4-7: Weather Profile for Summer Case Figure 4-8 shows the noise impact for the winter case. Figure 4-9 shows the noise 50

51 impact for the summer case. A comparison of the figures reveals that the shape of the contours in the winter case and the summer case are noticeably different. That is, in the winter case the higher intensity contours are significantly shorter and the lower intensity contours are narrower. This is the result of the better climb performance during winter, and suggests that if the wind is strong and the temperature is cool, it might be possible to turn prior to the Nahant peninsula. It is also interesting to note that the contours in the winter case are asymmetric. This may be explained by the fact that, in the winter case, the aircraft executes its turn to the right at a higher altitude and thus, because of the strong noise source directivity, the contours are shifted to and wider on the left than on the right. While the wind does tend to "push" the contour to the right, the lower wind strength in the winter case means that it does not have as much of an effect as in the summer case, where the contour is not very symmetric. This suggests that the location of a turn and the strength of the wind are important factors in the design of a procedure if the intent is to reduce the noise impact in all weather conditions. Table 4.1: Population impacted by single B767 departure from Runway 4R in winter and summer Scenario / db SEL 60 < < < < 100 > 100 Winter Case 332, ,046 67,871 9, Summer Case 329, ,826 48,261 12, Table 4.1 shows the number of people impacted by noise of different levels for each case predicted by NOISIM. It can be seen that the total number of people impacted by noise above 60 db is slightly lower for the summer than the winter despite the fact that the summer case shows higher values of noise exposure further along the flight path than the winter case. This difference is attributed to the fact that the winter flight track is closer to the shoreline, thus making up for the higher attenuation of noise. If the lateral component of both trajectories was the same, the noise impact in the winter case would be lower. The main point to remember, however, is that the shape of the contour in winter and summer is significantly different and that this 51

52 difference might enable a more radical change in the departure procedure when the weather conditions allow. That is, to turn before the Nahant peninsula. Figure 4-8: SEL noise contours for Winter Case Generated by NOISIM - Simulated Standard Noise Abatement Procedure at Boston Logan Runway 4R 52

53 Figure 4-9: SEL noise contours for Summer Case generated by NOISIM - Simulated Standard Noise Abatement Procedure at Boston Logan Runway 4R 53

54 54

55 Chapter 5 Weather-Specific Noise Abatement Procedure The results presented in Chapter 4 suggest that in certain weather conditions, aircraft will be able to gain sufficient altitude to execute the turn towards the Atlantic Ocean between the airport and the point on the extended centerline that is abeam of Nahant instead of after the aircraft passes Nahant as is done in the existing departure procedure. Based on this insight, a weather-specific departure procedure was developed for departures from Runway 4R. In this procedure, aircraft execute a turn to the right, from the runway heading of 040 to a heading of 090, at 500 ft above field elevation. The procedure is patterned on the departure procedure for Runway 22L (that is, for departures to the South from the same runway) where aircraft turn sharply to the left at 500 ft above field elevation. The goal of this procedure is to reduce the number of people impacted by noise by avoiding the coastal communities north of the airport and by partially avoiding the Nahant peninsula. To illustrate the feasibility and noise benefits of this change in procedure, both the existing procedure and the weather-specific procedure were flown by a transport rated pilot on Microsoft Flight Simulator 2000 for the conditions corresponding to the summer case in the previous section (the worse case in terms of climb performance). The aircraft trajectories were recorded and used as input to NOISIM. 55

56 Figure 5-1 shows the simulated trajectory for the existing procedure. As the figure shows, the aircraft, after making the heading change, flies parallel to the coastline and over the causeway connecting Nahant to Swampscott. The corresponding noise impact, shown in Figure 5-2, illustrates how the proximity of the aircraft to the coast results in significant noise impact in the communities of Lynn, Saugus and Swampscott that lie along the coast. 4 Repartre-1 - Fiht1Tmk Figure 5-1: Simulated trajectory for existing departure procedure - runway 4R, Boston Logan airport Figure 5-3 shows the simulated trajectory for the weather-specific procedure. The corresponding noise impact, shown in Figure 5-4, illustrates how the turn at 500 ft above field elevation directs most of the noise impact over water and significantly reduces the noise impact in the communities of Lynn, Saugus and Swampscott. Table 5.1 shows the number of residents impacted by noise in 10 db bins between 60 db and 100 db SEL. As the numbers in the table illustrate, the weather-specific procedure greatly reduces the number of people exposed to noise between 60 db and 90 db SEL. The results also indicate that the number of people impacted by noise greater than 60 dba is reduced by almost fifty-percent from 275,000 for the existing departure procedure to 140,000 for the weather-specific departure procedure. 56

57 Figure 5-2: SEL noise contours for existing departure procedure Table 5.1: Population impacted by noise (in 10 dba bins) specific departures from runway 4R for existing and weather- Scenarios / db SEL 60 < < < < 100 > 100 Existing Departure 405, ,083 43,825 5, Weather-Specific Dep. 293,745 94,574 18,882 5,

58 Figure 5-3: Simulated trajectory for weather-specific departure procedure - runway 4R, Boston Logan airport Figure 5-4: SEL noise contours for weather-specific departure procedure 58